[TSan][Darwin] Additional TSAN test requiring weak symbol for dyld64

[thetruestblue]

Summary

Tests require a weak symbol for dyld weak-def coalescing.

[thetruestblue]

@rsundahl do we think I should add weak attribute to OnPotentiallyBlockingRegionEnd as well? or leave as is?

[yln]

In D130732#3687617, @thetruestblue wrote:

@rsundahl do we think I should add weak attribute to OnPotentiallyBlockingRegionEnd as well? or leave as is?

Yes, it's the same mechanism (consistency), even if we don't need to make the test pass.

[thetruestblue]

This comment was removed by thetruestblue.

[thetruestblue]

Additional weak def

[rsundahl]

LGTM! Thanks for finding this one!

[rsundahl]

In D130732#3687617, @thetruestblue wrote:

@rsundahl do we think I should add weak attribute to OnPotentiallyBlockingRegionEnd as well? or leave as is?

I agree with @yln here @thetruestblue. In this case, better to be consistent.

compiler-rt/test/tsan/cxa_guard_acquire.cpp

 // RUN: %clangxx_tsan -O1 %s -o %t && %run %t 2>&1 | FileCheck %s
 
 #include <stdio.h>
 
 namespace __tsan {
 
+#if (__APPLE__)
+__attribute__((weak))
+#endif
 void OnPotentiallyBlockingRegionBegin() {
   printf("Enter __cxa_guard_acquire\n");
 }
 
+#if (__APPLE__)
+__attribute__((weak))
+#endif
 void OnPotentiallyBlockingRegionEnd() { printf("Exit __cxa_guard_acquire\n"); }
 
 } // namespace __tsan
 
 int main(int argc, char **argv) {
   // CHECK: Enter main
   printf("Enter main\n");
   // CHECK-NEXT: Enter __cxa_guard_acquire
   // CHECK-NEXT: Exit __cxa_guard_acquire
   static int s = argc;
   (void)s;
   // CHECK-NEXT: Exit main
   printf("Exit main\n");
   return 0;
 }

compiler-rt/test/tsan/default_options.cpp

 // RUN: %clang_tsan -O1 %s -o %t && %run %t 2>&1 | FileCheck %s
 #include <pthread.h>
 #include <stdio.h>
 
+#if (__APPLE__)
+__attribute__((weak))
+#endif
 extern "C" const char *__tsan_default_options() {
   return "report_bugs=0";
 }
 
 int Global;
 
 void *Thread1(void *x) {
   Global = 42;